Multiprocessor Support

With ColorSync version 2.5, the default ColorSync CMM supplied by Apple Computer and Linotype-Hell can take advantage of multiple processors. Multiprocessor support is transparent to your code--the CMM invokes it automatically if the required conditions are met.

When ColorSync Uses Multiple Processors

The default CMM takes advantage of multiprocessor support only if the following conditions are satisfied:

The MPLibrary was successfully loaded at boot time.
The CMM successfully links against the MPLibrary at runtime.
The number of processors available is greater than one.
The number of rows in the image is greater than the number of processors.
The source and destination buffers have the same number of bytes per row or have different locations in memory.

Unless all of these conditions are met, matching will proceed without acceleration. Multiprocessor support is currently supplied only for the following component request codes:

kCMMMatchBitMap
kCMMMatchPixMap

As a result, the default CMM invokes multiprocessor support only in response to the low-level CWMatchPixmap and CWMatchBitmap calls, or when those calls are invoked as a result of a call to the high-level QuickDraw matching routines (NCMBeginMatching and so on).

Efficiency of ColorSync's Multiprocessor Support

Depending on the image and other factors, ColorSync's matching algorithms take advantage of multiple processors with up to 95% efficiency (your mileage may vary). If you have two processors, for example, ColorSync can complete a matching operation in as little 53% of the time required by one processor. Additional processors will provide proportionally equal improvement.